If you already have a server access log, or can easily get one, start here. Upload a CSV or XLSX export, check the column mapping, and explore which recognised AI and search crawlers requested your pages.
Your log is read locally in your browser. It is not uploaded to this website or sent to an API. The analyser’s font and parsing libraries are bundled into the static HTML file, so it makes no third-party requests.
Use the analyser
- Export your access log as CSV or XLSX. A user-agent column is required; URL, status code, timestamp, and IP are optional.
- Select the file below. The tool recognises common column names such as
user_agent, http_user_agent, cs-uri-stem, and sc-status. - Review or override the suggested mappings, then choose Analyse log file.
- Explore the results and download a full annotated CSV or a per-bot summary.
Large files are easier to explore in a separate tab.
Open analyser in a new tab ↗
What you’ll see
- Summary: matched AI visits, AI platforms, pages crawled by detected bots, and assistant bot visits.
- Insights: assistant activity, failed AI requests, platform concentration, and training-crawler activity without assistant matches.
- Bot types: assistant, AI search, training, agent, and traditional search.
- Tables: top AI bots, top crawled pages, and response-code distribution.
- Detected visits: the first 500 rows are displayed; all matched rows are included in the full CSV export.
Downloads and privacy
The full CSV contains each detected request with its bot, company, type, user agent, URL, status, time, and IP where supplied. The summary CSV groups visits, unique pages, and errors by bot. Keep exported reports as carefully as the source logs, since they may contain IP addresses and private URLs.
You can also download the standalone HTML file and open it locally. It includes the parsers and font and works offline.
Download standalone HTML ↗
How to interpret the results
This version contains 96 detection patterns: 70 AI-related patterns and 26 traditional search-crawler patterns. These are pattern entries, not a count of verified or unique bots. The analyser uses the most specific matching pattern first.
A user agent is a label a client sends; it can be copied or spoofed. A detected visit does not prove a citation, recommendation, training use, or booking. Ordinary browser user agents cannot reliably identify an agentic browser from the user-agent string alone.
Log coverage depends on where the file came from. Origin logs may omit requests served entirely by a CDN or edge cache. Use edge logs too when you need that traffic. Raw text access logs need to be converted to CSV or XLSX first; the analyser reads the first worksheet of an Excel file.
No easy access to server logs?
The HOP AI Bot Monitor WordPress plugin provides ongoing monitoring of recognised bot requests that reach WordPress. Use it across your sites, then compare against server or edge logs when available.
